Identifying Significant Transcriptional Changes

Differential Gene Expression (DGE) analysis is a core transcriptomic technique designed to identify genes whose transcript levels significantly differ between experimental groups (e.g., control vs treated, healthy vs diseased).

Statistical Modeling of Count Data

RNA-seq read counts are discrete numbers exhibiting overdispersion, where variance exceeds the mean. Standard normal distribution models fail on RNA-seq counts. Bioinformatics packages like DESeq2 and edgeR employ negative binomial generalized linear models (GLMs) with empirical Bayes shrinkage to model variance accurately across small sample sizes.

Key Output Metrics

  • log2 Fold Change (log2FC): Indicates the magnitude and direction of expression change.
  • P-value & Adjusted P-value (FDR): The Benjamini-Hochberg adjustment controls false discovery rates across thousands of parallel hypothesis tests.